Engineering Letters, 24:1, EL_24_1_15
______________________________________________________________________________________
Modeling, Control and Optimization of a New
Swimming Microrobot Design
F. Srairi, L. Saidi, F. Djeffal and M. Meguellati
Abstract— This article deals with the study of a new swimming
microrobot behavior using an analytical investigation. The
analyzed microrobot is associated by a spherical head and hybrid
tail. The principle of modeling is based on solving of the coupled
elastic/fluidic problems between the hybrid tail’s deflections and
the running environment. In spite of the resulting nonlinear
model can be exploited to enhance both the sailing ability and
also can be controlled in viscous environment using nonlinear
control investigations. The applications of the micro-robot have
required the precision of control for targeting the running area in
terms of response time and tracking error. Due to these
limitations, the Flatness-ANFIS based control is used to ensure a
good control behavior in hazardous environment. Our control
investigation is coupled the differential flatness and adaptive
neuro-fuzzy inference techniques, in which the flatness is used to
planning the optimal trajectory and eliminate the nonlinearity
effects of the resulting model. In other hand, the neuro-fuzzy
inference technique is used to build the law of control technique
and minimize the dynamic error of tracking trajectory. In
particular, we deduct from a non linear model to an optimal
model of the design parameter’s using Multi-Objective genetic
algorithms (MOGAs). In addition, Computational fluid dynamics
modeling of the microrobot is also carried out to study the
produced thrust and velocity of the microrobot displacement
taking into account the fluid parameters. Our analytical results
have been validated by the recorded good agreement between the
numerical and analytical results.
Index Terms— ANFIS, Fuzzy computation, IPCM,
Flatness, Control, Microrobotics, Swimming.
I.
INTRODUCTION
N
owadays, the microrobots take the lion's share of the
different research fields of interest. They are increasingly
important necessities for precision applications due to their
speed in terms of response time and controllability in
dangerous environments [1-6]. The surgical assistance
microrobots are regularly used in operating rooms. Currently,
the researchers have developed miniature endoscopic capsule
robot able to explore intestinal conduits, arteries or veins, for
assist the surgeon in his diagnosis, based on the entirety of
Manuscript received 30 December, 2015.
F. Srairi is with Department of Electronics, University of Batna, 05000,
Algeria (e-mail:[email protected]).
L. Saidi is with Department of Electronics, University of Batna, 05000,
Algeria (e-mail:[email protected]).
F. Djeffal is with the Laboratory of Advanced Electronic, Department of
Electronics, LEPCM, University of Batna, 05000, Algeria (e-mail:
[email protected], [email protected])
m. meguellati is with the Department of Electronics, University of Batna,
05000, Algeria (e-mail: m_meguellati@ yahoo.fr).
submicron embedded sensors [7,8], or in military applications
such as the description of the nuclear area or define a large
moving targets [9-13].
Recently, many researches and practices are developed
including new materials and design aspects [1-5]. Most of
these microrobots used a huge number of parts, which leads to
microrobot behavior degradation in terms of complexity and
fabrication cost. In other hand, several investigations have
been conducted on microrobot systems, which are not based
on the biomimetic philosophy. New method was introduced by
Chang et al [1], in this method; an external magnetic field is
applied to rotate a small ferromagnetic screw in the liquid.
However, speed limitation is the main disadvantage of this
microrobot.
In this work, we have modeled a new swimming
microbobot design, which is compounded by a spherical head
and hybrid tail. The principle of modeling is based on solving
the coupled elastic/fluids problems between the hybrid tail’s
vibrations and running environment. Where the motion of the
hybrid tail is strongly created by an active joint made in
polymer ionic metal (IPCM) and the initial motion of the
microrobot is excited by a field magnetic outcome of the coil
set up at the head and occupied a half of the head's vacuum.
The use of coil in head allows us to ensure a better
controllability and stability of the microrobot device [14].
Despite of nonlinear model result from the analytical
modeling of our design and due to their ambiguous
applications, it is required fast and high strategy to achieve the
target. The control strategy presents a main topic in
microrobot field. Therefore, the aim of our control strategy is
to improve the performances of the device in running
environment, in terms of tracking trajectory and time
response. Our strategy is based on coupling between the
differential flatness and Neuro fuzzy inference (ANFIS)
techniques. In which, the use of flatness technique allows us to
eliminate the nonlinearity of the analytical model and
generates an optimal trajectory. In other hand, the ANFIS
strategy is used to build the law control to ensure a good
tracking trajectory and minimize the dynamic error basing on
back propagate method [15-16].
The analytical modeling is completed by computational
fluid dynamic (CFD) modeling because of their
complementary nature. Therefore, the use of CFD modeling
gives the kinematic patterns of the active joint. So, it’s taken
much time to capture some nonlinear and complex effects
such that vortices. This latter is ignored in analytical
modeling. The investigation is focused on one hybrid tail
attached to a body [1][17-21]. The numerical results indicate
that good agreement is achieved between the analytical
(Advance online publication: 29 February 2016)
Engineering Letters, 24:1, EL_24_1_15
______________________________________________________________________________________
modeling and CFD one in many points of kinematic patterns
of active joint.
The remainder of this work is organized as follows:
In section II we present the analytical model of swimming
microrobot. The control strategy is presented in section III.
The CFD modeling is discussed in section IV. In Section V,
we present the obtained results, highlighting, the comparison
between the analytical and CFD investigations and the
MOGA-based optimization is presented too. We provide
concluding and some remarks for future work in Section V.
II.
PRINCIPE OF THE MODELING
The structure of the investigated microrobot is illustrated
in fig.1 as it is shown from this figure, our microrobot design
is compound by a spherical head and a hybrid flexible tail.
The tail has two IPCM joints and two passive plates as links.
The plates are assumed to be rigid but thin and light. The
volume of the head is considered as a vacuum sphere for
placed the coil, drugs and sensors. The coil occupies 60% to
provide the initial driving force, and the remainder 40% of the
volume reserved for drugs and sensors.

r  i cos( )  j sin( )

w  i sin( )  j cos( )
Link
r/2

w

w
d 1 L1

j
Joint

w

w

w

w
B2

w

i
B1
w(z,t)
FB
α

rw

w

w
Figure 1. Cross sectional view of our proposed microrobot design with a
description of the tail.
The governing equation for the dynamic deflection
function w(X,t) of the tail performing flexural oscillations is :
 4W ( X ,t )
 2W ( X ,t )
EI

 F( X ,t )
(1)
4
X
X 2
where E presents the Young’s modulus, I is the moment of
inertial of the tail, µ is the mass per unit length of the Tail, F is
the external applied force per unit length, x is the spatial
coordinate along the length of the tail, and t is time. The
boundary conditions of (1) are the usual clamped and free end
conditions.

 W ( X ,t )
 2W ( X ,t )
 3W ( X , t


W ( x , t ) 
2

X

X
X 3

F ( X ,t )    w

 ( w ) f sin( ( t ))
(3)
4
where  w presents the density of fluid and  ( w ) is the
hydrodynamic function which is obtained from the solution of
Navier Stokes Equation:
Û  0 ,  P̂   2Û  iwû ,
(4)
Where Û presents the velocity field, P̂ is the pressure,  ,
 are the density and viscosity of the fluid, respectively.
Substituting (3) into (1) and rearranging we find:
 2W ( X ,t )
 4W ( X ,t ) A
a
  sin(  ( t ))
(5)
2
EI
X
X 4


With A    w  ( w ) f , a 
4
EI
Applying the above mentioned boundary conditions we
obtained the following expression of 2-D tail’s deflection
W(X, t):
1
 1

W ( X ,t )    w 2  Aw 2 t 2   0.1819 a  3.9  45 .4458 t
4
 2

1


 0.0249 X 2    0.0032 a  6.7284 ) X  A cos( 2tw )  ( 47 .8928 a 3 
8


1
1
1





 Exp  45 .6207  ( X  6.7543 .t )   0.6728 Xt  A
a
8
 a .94948 



 

 a 2 .4.8048 .10  4  a 2 .2.3930 .10  5  7.6361 .10 5
The lateral velocity of the tail can be calculated using 2-D
tail’s deflection W(X, t), as:
F1
U
cantilever tail moving in a fluid, the external applied load
F(X,t) can be calculate as:
)

 X 0

 W ( X ,t )
 2W ( X ,t )
 3W ( X , t ) 
 W ( x , t ) 


0

2
X
X
X 3

 X  L ,t  0
where L is the length of the tail (see Fig. 1). For a



U  



2
  w ( z ,t ) 
m

t


2
  w ( z ,t ) 
C D wS  m

z









(7)
The expression of Thrust force Fth is given by:
C  U 2S
(8)
Fth  D w
  mV M B
2
where  m is the magnetic ratio, m is the virtual mass
1
density at z=L expressed as m  S c2  w , with Sc presents
4
the width of the tail at the end z=L, S and CD are the wetted
surface area and drag coefficient, respectively.
III.
NUMERICAL MODELING
The CFD modeling allows us to solve the 3D Navierstocks equation using finite element method, where the flow is
considered laminar. The structure is modeled as a cylindrical
channel containing the microrobot design, to calculate the
unsteady solution the dynamic mesh is adopted, while the
microrobot’s volume creates other moving conditions with the
fluid. The all surfaces of the channel are defined as a wall with
no slip conditions and a mesh tetrahedral uniform and fine is
used to grid the structure. To make the precise of the solution
and close the reality the size of the channel is chosen as
(15x10-3 µm of radius and 30x10-3 µm), in which the volume
is mashed as 12, nodes 125, and faces 65 cells, where the
(Advance online publication: 29 February 2016)
Engineering Letters, 24:1, EL_24_1_15
______________________________________________________________________________________
fig. 2 illustrates the mesh distribution of the channel
containing the microrobot. The forces acting on the micro
robot are calculated with the sinusoidal excitation force of the
hybrid tail. In particular, the numerical data points should
stretch out by reason of building a numerical database and
compare it with the analytical results to validate the results
obtained. Therefore, to get an accurate to the results, we have
shown the independence of results and the channel size;
containing the micro robot and the mesh density by doing
varied the size of the channel and the mesh density, we have
spotted no notable differences between the computed thrust
force and velocity distribution.
Where Ii present the moment of inertia.
IV.
FLATNESS-ANFIS BASED CONTROL OF THE
MICROROBOT DESIGN
The feedforward controller is designed based on the
flatness input’s method. The flatness based controller produce
a flat input according to the given output signal which is work
on the controlled plant. The structure chart of our microrobot
with ANFIS-flatness control is shown in Fig. 4.
Flat trajectory
+
-
e
Fuzzification
Knowledge
Base
Defuzzification
Fat system
Δe
ANN
Y1, Y2
Figure 4. Flowchart of our proposed Flatness-ANFIS-based control.
Figure 2. Mesh of the channel containing the swimming microrobot.
A. Estimation of forces
Computational fluid dynamic allows calculating the
hydrodynamic force acting on each part of hybrid tail and the
distribution of velocity. Fig. 3 shows the different links of the
microrobot body and the parameters used to analysis the
hydrodynamic force. Where, Ci can be the center of mass and
center of inertia simultaneously, to evaluate the thrust force
generated by vibrating hybrid tail in fluid the following
recursive equations are used:
:
The head
Link 0
a0
b0
The tail
Link 1
a1

w
b1
c0
F1

 X  U 1 sin(  )  U 2 cos( )  Fdx
U 1  m cos(  )


(9)
FB
Z  U 1 cos( )  U 2 sin(  )  P  Fdz With 
U 2  
 
m

  U 2
mL

 


w
F1 c1
Center of mass
Figure 3. Definition of variables for force and moment analysis.
Fi  Fi 1  mi X  Gi
In witch mi is the mass of each link and Gi are the
localized force and the moment is calculated as:
M i  M i 1  I i   Gi
1
J
Where Fd is the drag force and equal to Fd  6rU ,
 V (    f )g
P
m
, V represents the total volume of the
microrobot, ρ is the density mass, g is the vector of the
gravity’s acceleration, j is the moment of inertia.
The geometrical and electomechanical configuration of the
simulated microrobot is given in table 1.
TABLE I
MICROROBOT DESIGN AND SIMULATION PARAMETERS
Parameters
M1

w
A. Model of the microrobot
In this section we present the forces acting on our
microrobot in fluid environment. So, the translational motion
of the microrobot is expressed by:
Values
Magnitude Attack angle (A)
60°
Pulsation ( w )

Viscosity dynamic (  )
0.008
Hydrodynamic function (  1 )
1
Robot’s Radius (r)
3.10-3 [ µm]
fluid density (  w )
1060[kg.m-3]
Magnetic ratio
0.8
Tail Length
4.10-3 [µm]
(Advance online publication: 29 February 2016)
Engineering Letters, 24:1, EL_24_1_15
______________________________________________________________________________________

B. Flatness-based control
By
Y2  Z 
choosing
1

the
flat
output Y1  X 
1

sin(  ) ,
cos(  ) , the flatness-based feedforward controller
becomes

Y1
1
 X  Y1 
2
 ( Y )  ( Y  P ) 2

1
2

Y2  g
1

Z  Y2  
( Y1 ) 2  ( Y2  P ) 2



  arctg Y1 
 Y  g 


 2




2 
Y1
Y1


U 1   d  Y1  1
2 



 ( Y )2  ( Y  P ) 2  ( Y )2  ( Y  P ) 2 
dt 
1
2
1
2









2 
Y2  g
Y2  g
1
 d 

  g 
 dt 2  Y2  


 ( Y )2  ( Y  P ) 2 
 )2  ( Y  P ) 2 
(
Y
1
2
1
2








2 



 Y 

1 d 
arctg 1  
U 2 


 dt 2 

 Y2  P  
(10)
C. Designing of ANFIS controller
The ANFIS structure considered as a multilayer
feedforward network, which was proposed initially as a
combination of fuzzy logic and artificial neural networks [15].
The ANFIS have both advantages of the neural network
learning, capability and the structured knowledge
representation employed in fuzzy inference systems. In the
pioneer work of Jang [17-18], it is demonstrated that ANFIS is
a universal competitive approximator compared to the most
other existing approaches. In order to keep the microrobot in
the desired trajectory, the strategy of the ANFIS control
consists to adjust in permanent the values of the corrector
gains. The neuro-fuzzy controller developed consists of two
inputs, error (e) and change of error ( de  e ) as it is shown in
Fig. 3.
Figure 5. Structure of ANFIS proposed for the microrobot’s control with two
inputs and one output.
The role of each layer of the network is described below:
 Layer 1: the output of each node gives the input
variable membership grade.
Layer 2: the firing strength associated with each
rule is calculated.
 Layer 3: the calculation of the relative weight of
each rule is achieved.
 Layer 4: the multiplication of normalized firing
strength by first order of Sugeno fuzzy
rule is realized.
 Layer 5: one node is composed and all inputs of
the node are added up.
The inputs to the ANFIS controller which are the input
error and the change in error are modeled by:
e( k )  Y1,2 Re f  Y1,2
(11)
e( k )  e( k )  e( k  1 )
The inputs of the fuzzy controller are error e and its
variance ratio ec, and u is the output. The universe of fuzzy
sets of e and u is {-1,-0.6,-0.2, 0, 0.2, 0.6, 1}, and the
corresponding fuzzy sets is {NB, NM, NS, ZE, PS, PM, PB}.
The universe of ec is {-1,-0.6,-0.2, 0, 0.2, 0.6, 1}, and its fuzzy
sets is {NB, NM, NS, ZE, PS, PM, PB}. The fuzzy rules can
be obtained by weighted average method. The fuzzy control
rules are shown in Table. 2.
TABLE II
FUZZY CONTROL RULES
E
ΔE
NB
NB
NM
NS
ZE
PS
PM
PB
NB
NB
NB
NB
NM
NS
ZE
NM
NB
NB
NM
NM
NS
ZE
PS
NS
NB
NM
NS
NS
ZE
PS
PM
ZE
NB
NM
NS
ZE
PS
PM
PB
PS
NM
NS
ZE
PS
PS
PM
PB
PM
NS
ZE
PS
PM
PM
PB
PB
PB
ZE
PS
PM
PB
PB
PB
PB
V. RESULTS AND DISCUSSION
In order to show the impact of our proposed design and to
ensure the best control of our microrobot device, the flatnessANFIS-based controller is investigated and developed, where
the results will be compared to the conventional approaches.
The typical data set should be stretching out as much as
possible in the entire input-output space of data, in order to
build an appropriate database for the ANFIS training set. The
data set used for the training of our fuzzy system is obtained
using the Matlab software. After the running of the learning
algorithm, we find that the triangular function has higher rate
of accuracy, where the recorded error for the training set
equals to 3.1x10-3, which is the square mean error of tracking
trajectory. It is to note that the number of epoch is set to 4000
and the method used for the train FIS is the back-propagation
algorithm. The comparison between the generated results
using different membership functions is investigated. The
ANFIS response surface of the partition of the input/output
variables using the best attained membership function is
illustrated in fig. 6.
Fig. 7 illustrates the evolution of the thrust force F as a
function of the radius of the microrobot’s head. The curves are
(Advance online publication: 29 February 2016)
Engineering Letters, 24:1, EL_24_1_15
______________________________________________________________________________________
plotted assuming a gradient magnetic field 80 mT.m-1. This
latter ensures the magnetization of the microrobot to reach the
saturation regime. Moreover, the thrust force is mainly
depended to the magnetization of the microrobot’s head. It is
20
Symbols: Numerical results
Thrust force for A=40°
Thrust force for A=45°
Thrust force for A=50°
lignes: Analytical results
18
Thrust force Fth [µN]
16
14
12
10
8
6
4
2
0
1.0
1.5
2.0
2.5
3.0
3.5
4.0
Time [s]
Figure 8. The thrust force as a function of the time.
18
16
Force F [n N]
14
1.5
2.0
2.5
3.0
20
18
B Permendur
B Carbon Steel 1010/1020
B NdFeB - 35
B Fe3O4
16
14
12
12
10
10
8
8
6
6
4
4
2
2
0
1.0
1.5
2.0
2.5
0
3.0
Radius of the head r [mm]
Figure 7. The thrust force F as a function of the radius of the head r for
different magnetic materials.
clearly observed that the radius of the head is increased with
the increase of the thrust force, and thus the controllability can
be improved.
Fig. 8 presents the variation of thrust force as a function of
time, when we know that the thrust force represents the
important parameter of push and guidance of the swimming
microrobot. So, the increasing of thrust force allows
improving the electromechanical performances in terms of
controllability and robustness for large wide of applications.
The variations of thrust force as a function of time shows that
the vibrating motion of hybrid tail is periodic. In other hand,
the amplitude of the excitation angle has a direct impact on the
values of thrust force. Thus, the increasing of angle excitation
leads to increase the thrust force when this latter is able push
the microrobot in forward and resist the vortices effects.
The width W [mm]
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0
0.55
60
50
Our design with L1=4mm
0.50
Our new design
Conventionnel [19]
0.45
40
0.40
30
0.35
0.30
20
The force Fm [nN]
1.0
20
The force Fm [n N]
Figure 6. ANFIS response surface.
Fig. 9 shows the thrust force as a function of the length L
and the width W of the tail compared with that of the
conventional case. It is easy to note that when the values of the
length and width are increased, the total thrust force can be
increased. This result leads to improve the device performance
compared to the conventional one. Therefore, our proposed
design provides better performance in comparison to that
given by the conventional structures, in terms of the thrust
force, velocity and controllability of the swimming
microrobot.
Fig.10 presents the behavior of the microrobot including
the trajectory tracking parameter using the proposed ANFISbased control. It is to note that an important tracking error
value is recorded for t=0s and t=0.5s. This period is
considered as transition time, for which the device behavior
should be improved. In this context, the obtained error is small
than other published result [19]. Moreover, the recoded
transition time provided by our control strategy is smaller than
the recorded values given by conventional approaches, which
means that our microrobot takes a short time to track the
desired trajectory.
0.25
10
0.20
0
5
10
15
20
25
The length L[mm]
Figure 9. The thrust force in function of the length L and the width W of the
tail compared with that of the conventional case.
(Advance online publication: 29 February 2016)
Engineering Letters, 24:1, EL_24_1_15
______________________________________________________________________________________
0.5
Error [mm]
0.4
Flat output Y1 [mm]
8
6
0.3
0.2
0.1
- Minimization of drag force FD (X )
0.0
4
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Time t [ms]
Desired trajectory
Mesured trajectory
2
0
- Minimization of weight Wa ( X )
Where X represents the input vector of design parameters,
which is given by: X  (r, L, B1, B2 ,  , ) .
3.5
-1
0
1
2
3
4
5
6
7
8
9
10
Flat output Y2 [mm]
Figure 10. The behavior of the microrobot in pursuit of the desired flat
trajectory of the computed torque control by using ANFIS control.
Figure 11. Velocity field for predominantly forward thrust force.
Fig.11 shows the distribution of velocity field along the
channel containing the swimming microrobot. It is clearly that
the high flow velocity appears to be concentrated around the
active joint because this latter presents the actuator of the
swimming microrobot. In the case of the hybrid tail’s
deflection is positive the velocity seems to be larger
concentrated around the end of the hybrid tail; where the
exciting force is physically accumulated and pushes forward
the microrobot. We note that the analytical model fails to
show the negative thrust force, but this is due to the interaction
between the hybrid tail and vortices induced by the deflection
of the hybrid tail. These impacts are not captured by the
Analytical calculations.
A. MOGA Optimization
Recently, most real engineering problems require the
simultaneous optimization of usually often competing
functions. The optimization algorithm is calculating the
solution of the objective function according to a common
optimization criterion between them. In most cases, the
objective functions are not well defined prior to the crucial
process. So, the whole problem will treated as a multi
objective function with non-measurable parameters. The aim
of MOGA optimization is simple to find and implement the
optimal solutions of multiple fitness functions in the research
space. The microrobot design can be optimized using MOGA
For the implementation of purposed MOGA-based
optimization of electromechanical parameters of the
swimming micorobot, programs and routines were developed
and used under Matlab 7.2 software, and the simulations are
carried out using Pentium IV 3-GHz 2-GB-RAM computer.
To select the better value, tournament selection is used for
parent database. Scattered crossover generates arbitrary binary
values. An optimization process was carried out for 20
population database and maximum number of generations is
taken to be 1080 generations, for which stabilization of the
fitness was achieved.
The optimal solution to our problem is presented in Pareto
Front, where Fig. 12 shows the surface of one. In which, each
point represents the optimal value of the fitness function
corresponding to the configuration of one design parameters.
In this case three points are selected and presented in TAB.3,
where each one is associated with the optimal fitness function.
Therefore, the configuration of swimming microrobot design
corresponding of each point is presented in TAB.4. It’s clearly
shown that these configurations indicate that each design
presents better performances, where the ratio between the
Fth
thrust force and the sum of weight and drag force
Wa  FD
has a value greater than one where, FD and wa are given by
FD  6rVr wa  V (    f )
It meant that the swimming microrobot able to resist the
vortices effect and has a better performances in terms of
stability and controllability.
Weight of microrobot Wa [N]
10
algorithm to improve the electromechanical performances. In
this case, three fitness functions can be studied and optimized.
Such that, thrust force (Fth), weight of microrobot (Wa) and
drag force (FD). The use of this investigation allows us to
establish an optimal design of the proposed microrobot, where
the following criteria will be respected:
-Maximization of thrust force Fth (X )
-0.5
-0.6
Design 3
-0.7
-0.8
-0.9
-1.0
-1.1
-1.2
0.00
Thr
u st
Design 2
0.02
forc
th
10
9
0.04
0.06
eF
11
Design 1
[uN
]
8
0.08
7
ag
Dr
F
ce
for
]
[nN
D
Figure 12. Pareto-optimal solutions of the swimming microrobot design.
(Advance online publication: 29 February 2016)
Engineering Letters, 24:1, EL_24_1_15
______________________________________________________________________________________
TABLE III
COMPARISON BETWEEN THE OPTIMIZED SWIMMING MICROROBOT AND
CONVENTIONAL DESIGN PARAMETERS
Symbol
r (µm)
L (µm)
B1 (µm)
B2(µm)
η(Pa.s-1)
ρ (Kg.m-3)
Thrust Forcee
Design 1
Design 2
Design 3
3.5x10-3
3.5 x10-3
1.7 x10-3
2.5 x10-3
0.008
1055
4.72×10-8
2.79 x10-3
4 x10-3
1.9 x10-3
2.2 x10-3
0.008
1055
3.82 ×10-8
3.10 x10-3
4.2 x10-3
1.7 x10-3
2.3 x10-3
0.007
1060
5.65 × 10-8
Conventional
design [5]
5 x10-3
10 x10-3
5 x10-3
2.0 x10-3
0.2
1060
1.316×10-9
TABLE IV
OPTIMIZED MICROROBOT DESIGN PARAMETERS
Objective function
Thrust force FTh
Weight Wa
Drag force FD
Design 1
4.72×10-8
-1.22×10-12
1.18×10-8
VI.
Design 2
3.82 ×10-8
-6.24×10-13
8.71×10-9
5.65 × 10-8
-8.55×10-13
9.54×10-9
CONCLUSION
The authors would like to thank the CDTA, Centre de
Développement des Technologies Avancées, staff for their
support and assistance (COMSOL).
REFERENCES
[3]
[6]
[7]
ACKNOWLEDGMENT
[2]
[5]
Design 3
In this paper, a new high performance modeling,
optimization and control of a new swimming Microrobot
Design have been presented. The analytical analysis has been
developed in order to evaluate the microrobot performances,
which are: the velocity and the total thrust force. The thrust
force, the velocity and microrobot geometry are considered as
important parameters for high controllability and reliability
performances. The proposed design and control strategy have
shown a high electromechanical performance in comparison to
the conventional structures. Moreover, the applicability of the
ANFIS-Flatness-based approach for the improvement of the
swimming microrobot design and its controllability has been
proved in this study. It can be concluded that the proposed
ANFIS-Flatness-based control is an efficient tool for high
performance microrobot-based applications. In addition, the
numerical model of our design has been developed to compare
and validate the analytical models, where the adequate in
several points exist between the predictions from the
analytical model and the numerical results. Moreover, the use
of the MOGA optimization in this work aims at the
optimization of swimming microrobot Capabilities. Our
investigations give hopeful results for future applications of
the microrobot device in many fields.
[1]
[4]
X. Chang, L. Zhang, X. He, “Numerical study of the thunniform mode
of fish swimming with different Reynolds number and caudal fin shape,”
J. Computers & Fluids, Vol. 68, pp. 54–70, 2012.
Abdul Kadir Muhammad, Shingo Okamoto, and Jae Hoon Lee,
"Computational Simulations and Experiments on Vibration Control of a
Flexible Two-link Manipulator Using a Piezoelectric Actuator,"
Engineering Letters, vol. 23, no.3, pp200-209, 2015.
M. Meguellati, F. Srairi, F. Djeffal and L. Saidi, "Control of a New
Swimming Microrobot Design Using Flatness-ANFIS-Based
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
Approach," Lecture Notes in Engineering and Computer Science:
Proceedings of The World Congress on Engineering 2015, WCE 2015,
1-3 July, 2015, London, U.K., pp 354-358.
H. Horiguchi, K. Imagawa, T. Hoshino, “Fabrication and Evaluation of
Reconstructed Cardiac Tissue and Its Application to Bio-actuated
Microdevices,” IEEE Trans on nanobioscience, Vol. 8, pp. 349-359,
2009.
L. Arcese, M. Fruchard, and A. Ferreira, “Endovascular Magnetically
Guided Robots: Navigation Modeling and Optimization,” IEEE Trans
on Biomedical Engineering , Vol. 59, pp. 977-987, 2012.
F. Srairi, M. Meguellati, L. Saidi and F. Djeffal, “Analytical Modeling
and Optimization of New Swimming Microrobot design using genetic
algorithm computations,” 14th IEEE International conference on
Sciences and Techniques of Automatic control & computer engineering,
STA'2013, pp. 265-268, Sousse, Tunisia, 2013.
F. Djeffal, D. Arar, N. Abdelmalek, M.A. Abdi, R. Mahamdi, A.
Errachid, “A junctionless-multigate design to improve the electrical
performances for deep submicron ISFET-based sensors,” Sensor Letters,
Vol. 9, pp. 2309-2311, 2011.
T. Bendib, F. Djeffal, M. A. Abdi, D. Arar, “An Analytical analysis of
the sensitivity behaviour for deep submicron ISFET sensors,” Third
International Conference Electrical Engineering Design, ICEEDT09,
Sousse, Tunisia, 2009.
M. Meguellati, F. Djeffal, D. Arar, F. Douak and L. Khettache, “New
RADFET Dosimeter Design For Radioactive Environment Monitoring
Applications,” Eng Lett, Vol. 20, EL 20-4-06, 2012.
M. Meguellati, F. Djeffal, “New Dual-Dielectric Gate All Around
(DDGAA) RADFET dosimeter design to improve the radiation
sensitivity,” Nuc Instr and Meth in Phys Res, Vol. A 683 pp. 24-28,
2012.
D. Arar , F. Djeffal , T. Bentrcia, M. Chahdi, “New junctionless
RADFET dosimeter design for low-cost radiation monitoring
applications,” Physica Status Solidi (C) Current Topics in Solid State
Physics Vol. 11, pp. 65-68, 2014.
F. Djeffal, M. Meguellati, “Multigate RADFET dosimeter for
radioactive environment monitoring applications,” Lecture Notes in
Electrical Engineering (LNEE), Vol. 229, pp. 301-313, 2013.
M. Meguellati, F. Djeffal, D. Arar, T. Bendib, L. Khettache, “RADFET
dosimeter design for environment monitoring applications,” Proceedings
of the International Conference on Microelectronics, ICM’2012, Algiers,
Algeria, pp. 1-4, 2012.
Dennis E. Curtin, Robert D. Lousenberg, Timothy J. Henry, Paul C.
Tangeman, Monica E. Tisack “Advanced materials for improved
PEMFC performance and life,” Journal of Power Sources Vol. 131, pp.
41-8, 2004. .
T. Bentrcia, F. Djeffal, D. Arar, M. Meguellati, “ANFIS-based
computation to study the nanoscale circuit including the hot-carrier and
quantum confinement,” The 5th International Conference on Modeling,
Simulation and Applied Optimization, pp. 1-5, 2013.
T. Bentrcia, F. Djeffal, E. Chebaaki, “ANFIS-based approach to study
the subthreshold swing behavior for nanoscale DG MOSFETs including
the interface trap effect,” The 24th International Conference on
Microelectronics, pp.1-4, 2012.
J. Jang, “ANFIS: adaptive-network-based fuzzy inference system,” J.
IEEE Transactions on Systems Man and Cybernytics, Vol. 23, pp. 665–
685, 1993.
F. Djeffal, S. Guessasma, A. Benhaya, M. Chahdi, “An analytical
approach based on neural computation to estimate the lifetime of deep
submicron MOSFETs,” Journal of Semiconductor Science and
Technology, Vol. 20, pp. 158-164, 2005.
L. Arcese, F. Matthieu, B. Felix, A. Ferreira, B. J. Nelson, “ Adaptive
Controller and Observer for a Magnetic Microrobot,” IEEE Transactions
on Robotics, Vol. 29, pp. 1060-1067, 2013.
Jin-liang Yan, and Zhi-yue Zhang, "Two-grid Methods for Characteristic
Finite Volume Element Approximations of Semi-linear Sobolev
Equations," Engineering Letters, vol. 23, no.3, pp189-199, 2015.
COMSOL Multiphysics, URL http://www.femlab.com.
Zhongda Tian, Shujiang Li, Yanhong Wang, and Quan Zhang, "Multi
Permanent Magnet Synchronous Motor Synchronization Control based
on Variable Universe Fuzzy PI Method," Engineering Letters, vol. 23,
no.3, pp180-188, 2015.
(Advance online publication: 29 February 2016)